Post on 11-Mar-2022
© MwAPATA Institute, 2021
Assessing Alternatives to Tobacco Farming for Smallholders in Malawi
Mrunal Shah, Jacob Ricker-Gilbert & Makaiko Khonje
MwAPATA Institute Working Paper No. 21/03
May 2021
Mr. Shah (shah123@purdue.edu) is a Research Assistant and PhD Candidate in Agricultural Economics at Purdue University. Dr. Ricker-Gilbert is an Associate Professor of Agricultural Economics at Purdue University. Dr. Khonje is a Research Fellow at the MwAPATA Institute.
P.O. Box 30883 Capital City, Lilongwe Malawi
Chilanga Drive, Off Blantyre Street, Area 10/44
MwAPATA Working Paper 21/03 i
Assessing Alternatives to Tobacco Farming for Smallholders in Malawi
Mrunal Shah, Jacob Ricker-Gilbert & Makaiko Khonje
Executive Summary
This study uses nationally representative data on smallholder farmers in Malawi collected
between 2009 and 2019, along with time series data on tobacco prices to consider the
profitability of Malawian tobacco and potential alternative crops such as groundnuts,
soybeans, pigeon pea and maize. We find that that the percentage of smallholder farmers
growing tobacco declined over the past 10 years from 16% in 2009/10 to 5% in 2018/19. This
is consistent with a decline in tobacco prices during that time. However, it is important to
note that the tobacco price decline occurred after a significant boom in the late 2000ʼs and
tobacco prices are still at or above the levels from 15-20 years ago. We find that smallholder
households who left tobacco cultivation shifted their land into maize cultivation and to a
lesser extent groundnuts and soybean. However, for the 5% of households who remain in
tobacco farming, the crop remains very profitable, generating far higher returns to land and
labor than do alternatives such as maize and grain legumes (soybean, groundnuts and
pigeon pea). The people who remained in tobacco cultivation expanded their area between
2009/10 and 2018/19 by 33% and their average yields increase by nearly 40%. This suggests
that the tobacco sub-sector in Malawi may have undergone something of a structural
transformation over the past decade, where inefficient producers have left as the price of
tobacco declined, while the most efficient producers remained. As such, our policy scenarios
clearly indicate that it will be very expensive to induce the remaining 5% of households to
move away from tobacco through a combination of taxes on tobacco and subsidies for other
crops. Therefore, we suggest that it would be more cost-effective for policy priorities in
Malawi to focus on improving value chains for other high value crops rather than directly
focusing on moving the remaining 5% of smallholder farmers out of tobacco cultivation.
Shah et al.
MwAPATA Working Paper 21/03 1
1. Introduction
Malawi is arguably the worldʼs most tobacco dependent economy. Sales of the crop account
for 50% of the nationʼs total export value, and Malawi is the top producer of burley tobacco
alongside Brazil and United States (United Nations, 2017). One of the unique feature of
Malawian tobacco is that smallholder farmers with less than a hectare are the main
producers, cultivating the crop as their main source of income (Otañez et al., 2007).
Despite the Malawian economyʼs dependence on tobacco, international donors and
health organizations, have long suggested that the Malawian government should help
smallholder farmers move out of tobacco. They cite the well-documented negative health
risks associated with consuming tobacco. In addition, there is growing evidence to suggest
that producing tobacco also poses direct risks to smallholder farmers themselves. For
example, Xia & Deininger (2019) found that children in Malawian communities with a higher
proportion of tobacco growing households spent more time engaged in casual labor and
were less likely to advance to next grade in school compared to children in communities
where tobacco cultivation was less prevalent. Additionally, they also found that children aged
five to fourteen who worked on tobacco farms were more likely to suffer from green tobacco
sickness, which was contracted by touching tobacco leaves. Health threats to tobacco
producers and consumers, along with concerns about child labor may justify the need for
alternatives to tobacco for Malawian farmers on public health and ethical grounds.
Despite these adverse health effects, farmers likely need a clear profit incentive to
move away from tobacco production on their own. Otherwise policy interventions such as
taxes and subsidies will be needed to change relative profitability and induce farmers to
leave tobacco and move to other crops. Soybean, groundnuts, pigeon pea and more intensive
maize cultivation are often advanced as potential alternatives to tobacco.
With these considerations in mind the objective of the present study is to estimate
profitability of tobacco and alternative cash crops for Malawian smallholders. We seek to
help answer the questions about the extent to which farmers can be expected to move away
from tobacco on their own and what incentives are needed to encourage this switch to
alternative crops. Specifically, we want to identify the price, cost and productivity scenarios
under which alternatives such as soybean, groundnuts, maize, and pigeon pea can rival
Shah et al.
MwAPATA Working Paper 21/03 2
tobaccoʼs profitability for farmers. We conduct this analysis using nationally representative
cross-sectional and panel datasets of smallholder farm households collected during the
2009/10, 2012/13, 2015/16, and 2018/19 growing seasons. First, we analyze the percentage
of farmers that switch in and out of tobacco cultivation over the years, and the demographic
characteristics of those who do so. Second, we examine the trends in export price and value
of Malawi tobacco over the years. Third, we estimate the cost of production and gross
margins for tobacco and potential alternative crops for Malawian smallholders. Finally, we
run a set of policy scenarios that adjust observed market conditions to reduce the
constraints that would make alternative crops more profitable for smallholders relative to
tobacco. Our analysis is intended to provide new farm-level information on the profitability
of tobacco and alternative crop cultivation over the past ten years in Malawi. The goal is that
this information will help policymakers and donors in Malawi better understand the barriers
smallholder farmers face to switch out of tobacco farming and assist in identifying the
potential incentives and mechanisms.
2. Data
Throughout this study, we use the Integrated Household Survey (IHS) and Integrated
Household Panel Survey (IHPS) datasets for Malawi. Both the datasets were collected by
the National Statistical Office of Malawi and supported by the World Bank Living Standards
Measurement Study (LSMS). Household-level data such as IHS and IHPS allow us to identify
potential heterogeneity in tobacco and alternative crop profitability at the farm level. The
cost of production analysis is based on the cross-sectional IHS dataset from 2009/10,
2015/16, and 2018/19 agricultural season. The cross-sectional data is representative of
smallholder farm households at both the national and district levels. In order to understand
the switching behavior of tobacco cultivating households we use the four waves of IHPS
panel dataset for 2009/10, 2012/13, 2015/16, and 2018/19 agricultural season. These panel
data are comprised of a sub-set of the households from the larger panel datasets. The panel
Table 1: Sample Size by Survey Wave for IHS and IHPS Datasets. Dataset 2009/10 2012/13 2015/16 2018/19
IHS 9,945 NA 9,455 8,768 IHPS 892 1,137 1,470 1,804
Shah et al.
MwAPATA Working Paper 21/03 3
Table 2. Switching in and out behavior of tobacco households. Households that grew tobacco in at least one of the four survey years
Percent of tobacco growing households
Always growing 16 Dropping 45
Adding 39 Total 100
Source: World Bank Integrated Household Panel Survey. Note: 20% of the households grew tobacco in at least one of four survey waves, these are defined as tobacco growing households; Always growing = household grew tobacco in all four seasons (2009/10, 2012/13, 2015/16, 2018/19); Dropping = household grew tobacco in 2009/10 and dropped it in a later season; Adding = household did not grow tobacco in 2009/10 and added it in a later season
households were tracked overtime, and households who split into multiple households were
all also tracked. This makes the panel data extremely useful for measuring household-level
changes in production and other decisions over time.
Table 1 shows the sample size across waves for IHS and IHPS datasets. The IHS cross-
sectional dataset interviewed 9,945 unique households in 2009/10, 9,455 in 2015/16 and
8,768 in 2018/19. The IHPS panel dataset interviewed 892 households in 2009/10, 1,137 in
2012/13, 1470 in 2015/16 and 1,804 in 2018/19. These households were tracked and
resurveyed over time. The number of households in the second, third, and fourth wave of the
IHPS dataset are higher than the first wave as some of the original household split over the
years, but these split-off households were also tracked over time. Population density data
used in this study is collected from Global High-Resolution Population Denominators
Project. The data on Malawian tobacco exports are collected from Tobacco Control
Commission (TCC).
3. Switching in and out of Tobacco Growing
In our sample, 20% of the farming households grew tobacco in at least one of the four survey
years in the IHPS dataset. We refer to these households as tobacco growing households.
Table 2 shows the switching in and out behavior of the tobacco growing households. We
found that just 16% of them cultivated tobacco in all the four survey waves. Additionally, 45%
of tobacco growing households dropped tobacco cultivation after 2009/10. Conversely, 39%
Shah et al.
MwAPATA Working Paper 21/03 4
Table 3: Characteristics of Always growing households and Dropping households in first survey wave 2009/10. (1) (2) (3) Always growing Dropping Difference (1) - (2)
Household size 6.17 5.26 0.913* (2.389) (2.297) (2.31) Total adults in HH 3.23 2.61 0.620*** (1.237) (1.196) (3.03) HH Head age (years) 45.74 40.14 5.608* (14.97) (13.12) (2.42) Female HH head = 1 0.11 0.06 0.0458 (0.312) (0.240) (1.04) HH head has some education = 1 0.23 0.19 0.0446 (0.428) (0.393) (0.65) Value of durable goods assets ('000 MWK) 254.57 71.63 182.9 (758.7) (127.4) (2.68) Death of HH member = 1 0.02 0.07 -0.0469 (0.146) (0.253) (-1.20) Land owned (ha) 1.29 0.90 0.392* (1.189) (0.629) (2.84) Herfindahl Diversity Index 0.42 0.44 -0.0175 (0.135) (0.122) (-0.82) Maize price (MWK/kg) 90.23 98.85 -8.625 (11.36) (48.94) (-1.19) Tobacco price (MWK/kg) 984.33 862.56 121.8 (498.9) (450.3) (1.55) Groundnut price (MWK/kg) 186.23 192.62 -6.387 (55.91) (68.20) (-0.58) Soybean price (MWK/kg) 137.66 141.15 -3.489 (15.16) (18.27) (-1.17) Pigeon Pea price (MWK/kg) 241.47 239.73 1.745 (8.259) (20.00) (0.58) Fertilizer price (MWK/kg) 424.38 4543.63 -4119.2 (252.5) (28104.3) (-1.00) HH borrowed on credit = 1 0.15 0.14 0.0126 (0.360) (0.344) (0.21) Land cultivated (ha) 1.63 1.07 0.560*** (1.081) (0.667) (4.14) Land rented for cultivation (ha) 0.30 0.23 0.0706 (0.883) (0.625) (0.59)
Continued on next page
Shah et al.
MwAPATA Working Paper 21/03 5
Table 3: Continued
(1) (2) (3) Always growing Dropping Difference (1) - (2) Hired labor (days) 14.64 9.76 4.881 (32.66) (26.95) (1.01) Maize land share 0.45 0.48 -0.0336
(0.170) (0.179) (-1.12) Tobacco land share 0.32 0.30 0.0180
(0.174) (0.164) (0.64) Groundnut Land share 0.13 0.11 0.0162
(0.144) (0.141) (0.67) Soybean Land Share 0.02 0.01 0.00794
(0.0559) (0.0424) (1.01) Pigeon pea Land Share 0.04 0.03 0.00450 (0.110) (0.0906) (0.28) Ganyu income (MWK) 16521.54 29691.06 -13169.5
(45270.0) (124449.9) (-0.71) Population density (Persons/sq.km) 279.56 226.80 52.76 (528.0) (110.6) (1.09) Distance to ADMARC (km) 6.48 6.47 0.0176 (3.603) (4.051) (0.03) Distance to Auction floor (km) 62.76 72.42 -9.655 (36.19) (47.65) (-1.26) Distance to border post (km) 33.56 33.24 0.310 (16.58) (14.95) (0.12) Distance to weekly market (km) 3.19 3.09 0.103 (4.292) (4.280) (0.14) Community has irrigation scheme = 1 0.06 0.09 -0.0271 (0.247) (0.289) (-0.57) Annual Mean Temperature (Deg C * 10) 206.24 210.78 -4.546* (11.55) (13.67) (-2.04) Annual Precipitation (mm) 932.43 986.36 -53.93 (144.0) (176.9) (-1.88) Observations 47 132 179
Source: World Bank Integrated Household Panel Survey data. Note: All calculations in 2019 MWK. Column (1) represents the baseline characteristics of households that cultivated tobacco in all four waves. Column (2) represents the baseline characteristics of households that cultivated tobacco in baseline (i.e., 2009/10) but dropped it at least one of the later waves. The t-test significance indicators for continuous variables were based on their inverse hyperbolic transformed (IHST) values. IHST values were used due to the skewed nature of underlying distributions.
Shah et al.
MwAPATA Working Paper 21/03 6
of tobacco growing households who did not grow tobacco in the first survey wave of 2009/10
added it in one or more of the subsequent survey rounds.
This suggest that there is i) movement into and out of tobacco cultivation depending on
market and growing conditions, and most people are not attached to growing tobacco
regardless of these conditions; ii) a larger share of households are dropping tobacco
cultivation compared to those adding tobacco.
Table 3 presents the comparison of characteristics between always growing households
who grew tobacco in each of the four survey rounds and dropping households who grew
tobacco during the 2009/19 season but subsequently stopped cultivating it. The table shows
that the two groups were statistically different from each other in 2009/10 on average in
several observable ways . The always growing households had a significantly higher number
of members on average at 6.17 compared to 5.26 for the dropping households. Specifically,
the always growing households had 3.23 adult members on average in 2009/10 compared to
2.61 for the dropping households. Older household heads were more likely to always grow
tobacco, while younger household head were more likely to drop tobacco. There is also
evidence that those who always grew tobacco had 0.39 hectares more land than those who
dropped tobacco after 2009/10 (a 43% difference). The same difference is true for land
cultivated in 2009/10 between always growing tobacco households and those who dropped
the crop after 2009/10. The former group cultivated 1.63 ha on average while the latter
cultivated 1.07 ha on average in 2009/10. Finally, there is some evidence to suggest that
those who always grew tobacco lived in climates that were about 0.5 degrees cooler on
average in 2009/10 than those who dropped tobacco. All of these descriptive results suggest
that land, and labor constraints are associated with people dropping tobacco and agro-
climatic conditions play some role as well. This is reflective of the land, labor and
environmental requirements to grow tobacco consistently over time.
Table 4 addresses the issue of what happens to tobacco land after people stop
growing the crop. It is interesting to note that on average, maize received the highest share
of land that was reallocated away from tobacco in all three years. Legume crops such as
groundnuts, common beans, soybeans and pigeon pea all received an increase in land that
was re-allocated to maize, but the increase was smaller than for maize. Another finding that
Table 4: Share of land cultivated for various crops amongst households that grow and those that stop growing tobacco 2012/13 2015/16 2018/19
HHs that stopped growing tob. in 2012/13 but grew it in
2009/10
HHs that grew tob.
in 2012/13
(1) - (2)
HHs that stopped growing tob. in 2015/16 but grew it in
2012/13
HHs that grew
tob. in 2015/16
(4) - (5)
HHs that stopped growing tob. in
2018/19 but grew it in 2015/16
HHs that grew
tob. in 2018/19
(7) - (8)
(1) (2) (3) (4) (5) (6) (7) (8) (9) --Share of land cultivated-- --Share of land cultivated-- --Share of land cultivated--
TOBACCO 0% 25% -25% 0% 33% -33% 0% 31% -31% MAIZE 57% 39% 19% 72% 46% 26% 49% 35% 14% GROUNDNUT 18% 11% 7% 7% 6% 1% 13% 7% 5% COMMON BEANS 3% 4% -1% 2% 2% 0% 6% 0% 2% SOYABEAN 4% 4% 1% 5% 4% 2% 9% 6% 3% PIGEONPEA 6% 6% 0% 5% 2% 2% 5% 5% -1% NKHWANI 2% 6% -3% 4% 3% 0% 14% 8% 6% RICE 0% 0% 0% 0% 0% 0% 0% 0% 0% GROUND BEAN 0% 0% 0% 0% 0% 0% 0% 0% 0% SWEET POTATO 1% 0% 0% 1% 1% 0% 1% 1% 0% IRISH POTATO 1% 0% 0% 0% 0% 0% 0% 1% 0% FINGER MILLET 0% 0% 0% 0% 0% 0% 0% 0% 0% SORGHUM 1% 2% 0% 2% 2% 1% 1% 1% 0% PEARL MILLET 0% 0% 0% 0% 0% 0% 0% 0% 0% COTTON 2% 0% 1% 1% 0% 1% 0% 0% 0% SUNFLOWER 1% 0% 0% 0% 0% 0% 0% 0% 0% SUGAR CANE 0% 0% 0% 0% 0% 0% 0% 0% 0% CABBAGE 0% 0% 0% 0% 0% 0% 0% 0% 0% TANAPOSI 0% 0% 0% 0% 0% 0% 0% 0% 0% THERERE/OKRA 0% 0% 0% 0% 0% 0% 0% 0% 0% TOMATO 0% 0% 0% 0% 0% 0% 0% 0% 0% ONION 0% 0% 0% 0% 0% 0% 0% 0% 0% PEAS 2% 1% 0% 0% 0% 0% 0% 0% 0% PAPRIKA 0% 0% 0% 0% 0% 0% 0% 0% 0%
Source: World Bank Integrated Household Panel Survey
Shah
et a
l.
7 M
wAP
ATA
Wor
king
Pap
er 2
1/03
MwAPATA Working Paper 21/03 8
Table 5: Percentage of households that did not participate in crop cultivation. 2012/13 2015/16 2018/19 2018/19 HHs that
stopped growing tob. in
2012/13 but grew it in 2009/10
HHs that stopped
growing tob. in 2015/16 but
grew it in 2012/13
HHs that stopped
growing tob. in 2018/19 but
grew it in 2015/16
HHs that grew tob. in
all four survey years.
No. of households 126 115 158 53 (i) % of HHs that did not grow crops 10 16 16 0
% of (i) that practiced livestock farming 33 28 23 92
% of (i) that ran household enterprise 33 44 31 34
% of (i) with at least one member working as an employee
50 39 46 9
% of (i) with at least one member engaged in casual labor (ganyu)
50 72 62 60
Source: World Bank Integrated Household Panel Survey
bears more exploration is the rapid increase in share of area planted to the vegetable Nkhwani over time for both tobacco growing households and those who left tobacco. This has implications for both household nutrition, income and food security.
Table 5 indicates that in 2013 about 10% of households who stopped growing tobacco exited crop cultivation altogether, while 16% did so in 2016 and another 16% did so in 2019. The table further shows what these households who exited tobacco did for a living in those years. Majority of these households, who stopped growing tobacco and other crops, had at least one member engaged in ganyu (casual labor) or worked as an employee. About one-third of these households ran a household enterprise and one-fourth practiced livestock farming. This suggests a mixed picture of the outcomes for those who stopped growing tobacco and left crop cultivation altogether. A positive sign of economic transformation could be seen in the fact that a significant number of people formerly growing tobacco are now engaged in businesses or works as employees in other activities. However, the fact that the most common occupation for those who left tobacco and crop cultivation altogether was to
Table 6: Characteristics of households that stop growing tobacco and households that grow tobacco. 2012/13 2015/16 2018/19
HHs that stopped growing tob. in
2012/13 but grew it in 2009/10
HHs that grew tob.
in 2012/13 (1) - (2)
HHs that stopped growing tob. in
2015/16 but grew it in 2012/13
HHs that grew tob.
in 2015/16 (4) - (5)
HHs that stopped growing tob. in
2018/19 but grew it in 2015/16
HHs that grew tob.
in 2018/19 (7) - (8)
(1) (2) (3) (4) (5) (6) (7) (8) (9) Household size 4.88 5.69 -0.813** 4.82 5.39 -0.566* 4.28 5.44 -1.153***
(2.062) (2.374) (-3.05) (2.351) (2.443) (-1.98) (2.041) (2.276) (-4.94)
Total adults in HH 2.64 3.22 -0.581*** 3.00 3.20 -0.203 2.66 3.23 -0.571***
(1.110) (1.365) (-3.87) (1.499) (1.559) (-1.11) (1.276) (1.407) (-3.94)
HH Head age (years) 38.42 44.00 -5.576*** 41.16 44.07 -2.917* 38.65 44.97 -6.323***
(13.81) (14.43) (-3.31) (15.32) (14.55) (-1.65) (14.69) (14.06) (-4.10)
Female HH head (yes= 1) 0.12 0.07 0.0525 0.18 0.12 0.0578 0.15 0.13 0.0269
(0.326) (0.252) (1.54) (0.382) (0.323) (1.40) (0.360) (0.332) (0.73)
HH head has some education (yes= 1)
0.22 0.23 -0.0110 0.21 0.25 -0.0355 0.41 0.32 0.0937
(0.413) (0.420) (-0.22) (0.409) (0.432) (-0.70) (0.494) (0.467) (1.82)
Value of durable goods assets ('000 MWK)
81.71 233.97 -152.3* 272.55 171.13 101.4*** 106.76 242.70 -135.9*** (147.0) (937.7) (-1.80) (1115.0) (452.3) (1.10) (375.3) (866.3) (-1.84)
Death of HH member (yes= 1) 0.07 0.07 -0.00162 0.26 0.14 0.124** 0.11 0.11 -0.00178
(0.260) (0.262) (-0.05) (0.442) (0.347) (2.71) (0.311) (0.313) (-0.05)
Land owned (ha) 0.65 1.06 -0.410*** 0.45 0.83 -0.374*** 0.36 0.78 -0.423***
(0.628) (0.786) (-4.78) (0.802) (0.912) (-3.61) (0.510) (0.887) (-5.32)
Herfindahl Diversity Index 0.51 0.38 0.136*** 0.58 0.46 0.123*** 0.41 0.38 0.0224
(0.271) (0.140) (5.51) (0.338) (0.155) (4.30) (0.258) (0.167) (0.98)
Maize price (MWK/kg) 204.35 209.80 -5.443 254.29 250.13 4.158 32.20 33.20 -0.997
(33.44) (30.91) (-1.43) (27.84) (35.49) (1.07) 134.76 138.94 -4.172
Continued on next page
9 M
wAP
ATA
Wor
king
Pap
er 2
1/03
Shah
et a
l.
Tobacco price (MWK/kg) 970.80 1167.83 -197.0 536.45 631.60 -95.16 (32.59) (34.43) (-1.16)
(213.8) (615.3) (-3.42) (148.0) (404.2) (-2.42) 543.46 574.89 -31.43
Groundnut price (MWK/kg) 269.79 273.46 -3.678 352.01 343.11 8.900 (188.4) (326.7) (-1.07)
(43.94) (58.73) (-0.59) (106.6) (109.9) (0.69) 178.52 177.74 0.786
Soybean price (MWK/kg) 301.37 313.32 -11.95 357.46 370.79 -13.33 (49.96) (54.69) (0.14)
(51.48) (76.56) (-1.50) (78.51) (113.7) (-1.10) 201.73 195.42 6.316
Pigeon Pea price (MWK/kg) 246.28 248.69 -2.412 483.64 490.97 -7.333 (39.03) (34.69) (1.60)
(19.13) (25.87) (-0.88) (56.50) (61.82) (-1.03) 245.56 252.29 -6.736
Fertilizer price (MWK/kg) 831.45 15546.30 -14714.9 483.47 493.36 -9.886 (32.29) (38.00) (-1.76)
(2659.5) (93729.2) (-1.75) (106.5) (147.4) (-0.62) 403.12 388.95 14.17
HH borrowed on credit = 1 0.13 0.22 -0.0929* 0.23 0.30 -0.0767 (120.4) (125.7) (1.07)
(0.335) (0.416) (-2.04) (0.421) (0.462) (-1.45) (0.446) (0.469) (-1.03)
Land cultivated (ha) 0.76 1.36 -0.608*** 0.77 1.29 -0.519*** 0.61 1.35 -0.734***
(0.597) (0.783) (-7.22) (0.688) (0.777) (-5.87) (0.533) (0.906) (-8.98)
Maize land share 0.57 0.39 0.187*** 0.72 0.46 0.258*** 0.49 0.35 0.138***
(0.262) (0.193) (6.84) (0.262) (0.205) (9.12) (0.233) (0.190) (5.84)
Tobacco land share 0.00 0.25 -0.248*** 0.00 0.33 -0.334*** 0.00 0.31 -0.308***
(0) (0.179) (-14.84) (0) (0.212) (-15.51) (0) (0.219) (-16.14)
Groundnut Land share 0.18 0.11 0.0657** 0.07 0.06 0.00715 0.13 0.07 0.0550***
(0.220) (0.134) (3.08) (0.164) (0.111) (0.43) (0.177) (0.109) (3.45)
Soybean Land Share 0.04 0.04 0.00661 0.05 0.04 0.0167 0.09 0.06 0.0348
(0.106) (0.0812) (0.59) (0.138) (0.103) (1.15) (0.206) (0.116) (1.94)
Pigeon pea Land Share 0.06 0.06 0.00223 0.05 0.02 0.0232 0.05 0.05 -0.00564
(0.147) (0.139) (0.13) (0.129) (0.0709) (1.96) (0.120) (0.112) (-0.43)
Ganyu income (MWK) 60,735.58 53,678.65 7056.9 96,126.48 124,177.6 -28051.1 191,281.6 116,072.1 75209.5
(150393.5) (157217.5) (0.38) (164241.4) (433201.9) (-0.66) (580877.3) (224465.1) (1.65)
Distance to ADMARC (km) 5.94 7.50 -1.559*** 8.22 6.33 1.888** 6.13 6.24 -0.112
(3.964) (4.051) (-3.27) (4.682) (3.784) (3.83) (4.830) (4.523) (-0.22)
Observations 125 163 288 114 187 301 158 192 350
Source: World Bank Integrated Household Panel Survey data. Note: All calculations in 2019 MWK. The t-test significance indicators for continuous variables were based on their inverse hyperbolic transformed (IHST) values. IHST values were used due to the skewed nature of underlying distributions
10
Mw
APAT
A W
orki
ng P
aper
21/
03
Shah
et a
l.
Shah et al.
MwAPATA Working Paper 21/03 11
work as a ganyu laborer on another farm, raises questions about the well-being of these households as ganyu is generally considered an income source of last resort (Alwang & Siegel, 1999; Ricker-Gilbert et al., 2014).
Table 6 provides an ex-post comparison of the characteristics of households that stopped growing tobacco and households that grew tobacco. The results from table 6 indicate that households that stop growing tobacco dedicate a greater share of their land to maize and ground nut compared to the households that grew tobacco. Furthermore, table 6 echoes the ex-ante characteristic comparison presented in table 3 demonstrating the land, and labor constraints associated with households leaving tobacco cultivation.
4. Price trends for tobacco and alternatives over time
Figure 1 shows the real export price of Malawian tobacco in 2019 US Dollar per kg. The real export price of Malawian tobacco hovered around USD 3.00 per kilogram from 2003 to 2006 when it increased tremendously to nearly USD 5.00 per kilogram during the global commodity boom of the late 2000ʼs. The prices since 2010 have experienced relative declines back towards their levels in the early 2000ʼs, but still remain equal to or higher than average prices 15-20 years ago.
Figure 1: Real price for tobacco exported from Malawi to World market in 2019 US $ per kg
Source: Tobacco Control Commission. Note: Real prices generated by deflating with US CPI values from World Bank data for CPTOTSAXN indicator
3.143.38
2.98 2.983.28
4.924.47
4.84
3.86 3.83
4.52 4.40 4.30
3.943.77 3.76
3.253.43
2
2.5
3
3.5
4
4.5
5
5.5
6
200320042005200620072008200920102011201220132014201520162017201820192020
2019
usd
/kg
Shah et al.
MwAPATA Working Paper 21/03 12
Figure 2: Real Gross Export Value for Malawian Tobacco in 2019 Kwacha
Source: Tobacco Control Commission. Note: Real prices generated by deflating with US CPI values from World Bank data for CPTOTSAXN indicator.
Figure 2 shows the gross export value of Tobacco exported from Malawi to world market in 2019 Malawi Kwacha. We found that the gross export value in 2020 had declined by 60% since 2010. However, the recent declining tobacco export values are still above their value in early 2000ʼs. The trend in export value over the past 20 years is very similar to the trend in tobacco prices during that time period, as seen earlier in Figure 1.
Figure 3 shows the trends in real average farmgate tobacco prices received by smallholder over three seasons in 2019 US $ per kilogram from the IHS dataset. Almost all of the smallholders in the sample who grew tobacco grew the burley variety rather than flue cured. Average farmgate tobacco prices decreased between 2009/10 and 2018/19. This is consistent with the trend in tobacco prices and export values between 2010 and 2019 as seen in Figure 1 and Figure 2, suggesting a decline in tobacco over the past decade.
While the trend in farm gate tobacco prices is informative, in order to understand farmers decision-making processes tobacco prices should be compared to those of other crops. Figure 4 shows the ratio of average farmgate prices of alternative crops (maize, groundnut, soybean and pigeon pea) relative to tobacco using the IHS datasets. Results show that in
286321 305 313
401
518
597
804
548
605
396
417
451
513
507
434
379
322
0
100
200
300
400
500
600
700
800
900
20032004
20052006
20072008
20092010
20112012
20132014
20152016
20172018
20192020
2019
Mill
ion
$
Shah et al.
MwAPATA Working Paper 21/03 13
Figure 3: Real Average farmgate tobacco prices in 2019 US $ per kg
Figure 4: Real output farmgate price ratios (crop price / tobacco price) in 2019 Kwacha
Source: World Bank IHS data Source: World Bank IHS data
2009/10 all four alternative crops were priced between 10-30% of tobacco on a per kg basis. The relative price ratios for alternative crops improved relative to tobacco in 2015/16 at 30% for maize, 50% for groundnut, 50% for soybean and 70% for pigeon pea. This likely reflected higher prices for these food crops during a drought year. Regardless it is important to note that tobacco prices and yields still remained higher even under these less favorable conditions. In 2018/19 the relative price ratios dropped back to 20% for maize, 30% for soybean and groundnut and 30% for pigeon pea compared to tobacco.
5. Cost of production and returns to tobacco and alternative crops.
Table 7 shows descriptive results from the IHS repeated cross sectional data on the number and percentages of households growing tobacco in Malawi and the changes in area under cultivation, total output and yields. This table reveals some very important information. First, in row iv) we see that the percentage of households growing tobacco nationally declined from 15% in 2009/10 to 6% in 2015/16 and to 5% 2018/19. As such the area under tobacco in row v) declined from 147,000 hectares in 2009/10 to 82,000 hectares in 2015/16 but rose to 92,000 hectares in 2018/19.
1.02
0.92
0.86
0.75
0.80
0.85
0.90
0.95
1.00
1.05
2009/10 2015/16 2018/19
USD
/ KG
1.0 1.0 1.0
0.1
0.3
0.2
0.3
0.5
0.3
0.2
0.5
0.30.3
0.7
0.3
0.0
0.2
0.4
0.6
0.8
1.0
2009/10 2015/16 2018/19
Crop
Pric
e/To
bacc
o Pr
ice
Tob. Maize G. Nut S. Bean P. Pea
Shah et al.
MwAPATA Working Paper 21/03 14
Table 7: Tobacco participation and production by smallholders over time
However, despite the decline in participation rates and total acreage over time, tobacco output in row vi) only declined from 130,000 tons in 2009/10 to 100,000 tons in 2015/16 and rose back up to 120,000 tons 2018/19. It is also interesting to note that the average area under tobacco for those who grew tobacco increased consistently from 0.40 hectares in 2009/10 to 0.53 hectares in 2018/19, a 33% increase. At the same time average tobacco yields for those who grew tobacco increased consistently for those who grew tobacco over time, from 984 kilograms/hectare in 2009/10, to 1,281 kilograms/hectare in 2015/16, to 1,372 kilograms/hectare in 2018/19, nearly a 40% increase in yields.
The fact that participation rates in tobacco dropped substantially overtime, and total tobacco area cultivated dropped, while at the same time area cultivated and yields increased for those who remained in tobacco cultivation suggests that the remaining 5% of the smallholder population that grew tobacco in 2018/19 were the most efficient producers of the crop. They were able to increase yields between 2009/10 and 2015/16 even though the latter year was a drought year. This is important to consider when estimating cost of production for tobacco and other crops, based on the information provided by households in the IHS dataset.
Table 8 presents estimates of the relative returns to land, labor through corresponding cost and profitability build-ups for tobacco, maize, groundnuts, soybeans, and pigeon pea in 2009/10, 2015/16 and 2018/19. While the first row of table 8 show that the percentage of smallholders cultivating tobacco declined over time, it also clearly shows that the percentage of smallholders cultivating groundnut fluctuated across seasons. The percentage of farmers growing groundnuts declined between 2009/10 and 2015/16 and increased between
2009/10 2015/16 2018/19 i) Number of households growing tobacco 1,335 561 440 ii) Estimated number of smallholders growing tobacco (weighted) 368,038 183,774 174,370 iii) Total number of households (weighted) 2,515,879 2,900,000 3,236,919 iv) Percent of households growing tobacco (weighted) 15% 6% 5% v) Estimated total area under tobacco in ʻ000 ha (weighted) 147 82 92 vi) Estimated total tobacco output in ʻ000 tons (weighted) 130 100 120 vii) Average Area under tobacco if cultivating (ha) 0.40 0.45 0.53 viii) Yields (kg/ha) (mean) 984 1,281 1,372 Source: World Bank Integrated Household Survey
Table 8: Cost of Production: Mean return to labor, land and corresponding cost build up 2009/10 2015/16 2018/19
Tob. Maize G. Nut S. Bean P. Pea Tob. Maize G. Nut S. Bean P. Pea Tob. Maize G. Nut S. Bean P. Pea
Percent of hhs growing crop 15 95 26 5 21 6 92 15 10 22 5 93 27 16 29
Area (ha) 0.4 0.43 0.25 0.26 0.2 0.45 0.35 0.27 0.28 0.14 0.53 0.29 0.25 0.29 0.13
Est. total area (000ʼs ha) 147 1,022 162 36 106 82 924 120 81 86 92 883 212 146 119
Revenue Yield (kg/ha) 984 1959 743 749 613 1281 1841 943 849 668 1372 2198 1060 995 837
Total Output (kg) 351 662 155 167 99 567 515 228 219 74 697 526 216 258 82
Est. total output (mill. tons) 0.13 1.59 0.1 0.02 0.05 0.1 1.38 0.1 0.06 0.05 0.12 1.58 0.19 0.13 0.08
Gross Revenue (000's MWK/ha) 726 218 161 109 139 949 409 327 285 304 935 322 223 197 178 Costs (000's MWK/ha)
Seed 0.5 3.9 2.7 1.2 1.8 5.4 11.3 10.2 7.4 7.1 15.2 11.1 9.3 6.4 4.0
Seed Transport 0.1 0.1 0.1 0.0 0.0 0.3 0.3 0.2 0.0 0.1 0.9 0.4 0.2 0.2 0.1
Fertilizer 49.5 13.5 2.4 2.8 10.4 109.1 27.6 5.2 8.4 20.8 114.9 25.2 5.0 7.6 21.6
Fertilizer Transport 1.1 0.5 0.1 0.1 0.4 2.1 0.7 0.1 0.1 0.5 2.8 0.9 0.1 0.2 0.7
Hired Labor 4.3 3.6 3.4 2.6 4.0 10.6 5.9 8.0 6.4 4.9 22.2 6.4 6.8 6.6 6.1
Transport to buyer 30.5 0.0 0.2 0.5 0.1 27.8 0.0 0.8 0.6 0.0 24.0 0.0 0.5 0.6 0.0
Land 26.7 22.2 24.1 26.9 25.8 33.4 33.8 29.7 29.4 35.3 31.1 35.1 29.3 30.0 35.4
HH labor 281.1 174.7 199.1 191.7 182.9 350.4 231.3 264.2 228.5 250.0 397.6 288.5 350.4 276.6 331.7
Total costs exclʼ land 367.0 196.3 208.0 198.9 199.5 505.9 277.1 288.8 251.4 283.4 577.6 332.5 372.3 298.1 364.3 Total costs exclʼ hh labor 112.6 43.8 33.1 34.1 42.4 188.8 79.6 54.3 52.4 68.7 211.1 79.1 51.2 51.5 67.9 HH labor days (days/ha) 316 194 236 218 219 333 215 247 210 262 272 223 253 203 280
Returns to land (000's MWK/ha) 359.3 22.0 -47.1 -90.3 -60.4 443.4 131.6 38.4 33.1 20.9 357.1 -10.9 -149.0 -100.6 -186.6 Return to labor (000's MWK/day) 5.0 1.7 1.1 0.6 0.5 5.6 3.7 2.1 2.3 1.8 8.7 2.2 1.4 1.4 0.5 Source: World Bank Integrated Household Survey data. Note: All calculations in 2019 MWK. Figures computed using weights to ensure representativeness. Tob.=Tobacco; G.Nut=Groundnut; S. Bean=Soya bean; P.Pea=Pigeon Pea.
Mw
APAT
A W
orki
ng P
aper
21/
03
16
Shah
et a
l.
Shah et al.
MwAPATA Working Paper 21/03 16
2015/16 and 2018/19. Furthermore, the percentage of farmers growing pigeon pea increased over the years from 21% in 2009/10 to 29% in 2018/19. Additionally, percentage of smallholders in the sample who grew maize declined over the years from 95% in 2009/10 to 93 % in 2018/19 while the percentage of farmers growing soybean increased from 5% in 2009/10 to 16% in 2018/19. Average area1 devoted to groundnut decreased between 2009/10 and 2015/16 but increased in 2018/19. Average area dedicated to pigeon pea declined over the years from 0.2 ha in 2009/10 to 0.13 ha in 2018/19 about 35% decline. Conversely, average area dedicated to soybean increased by 12% between 2008/19. Similar to pigeon pea, average area devoted to maize declined over the years from 0.43 ha in 2009/10 to 0.29 ha in 2018/19.
Table 8 also shows that tobacco farmers experienced substantially higher yields for their tobacco than farmers did for other cash crops. This is consistent across seasons and holds even during the drought season of 2015/16. While the other cash crops had lower yields, their relative yield rank compared to tobacco was consistent over time. Groundnut yields came in second to tobacco, soybean yields came in third, and pigeon pea yields came in last. While maize enjoyed higher yields than tobacco across all years, the yield advantage of maize has declined over the years. Related to yields and prices, table 8 also presents gross revenue, costs and returns to land and labor for tobacco and potential alternative cash crops. To that end tobacco farmers who cultivated the crop obtained higher gross revenues2 on average than did farmers who cultivated other alternative crops. Meanwhile, the costs for tobacco are also higher than that of other alternatives. Fertilizer costs for tobacco are almost four times that of maize which has the second highest fertilizer costs. Furthermore, tobacco cultivation incurred highest hired labor costs and own household labor costs (opportunity
1 We used the GPS values for plot area. The missing plot areas were imputed with the self-reported values.
We divide the plot area among respective crops based on steps provided by Evans School Policy Analysis &
Research Group (EPAR).
2 Gross revenues were generated using quantity of crop which was both sold and unsold. The value of unsold
crop was based on the prices received for sold crop by the household. If the household did not sell the crop,
then the median price at community level were used (only if there were more than 7 observations). Similarly,
if community level prices were not available then district level median prices were used (again if more than 7
observations were available) and lastly if district prices were not available then national median prices were
used.
Shah et al.
MwAPATA Working Paper 21/03 17
Figure 5: Return to land sensitivity analysis.
Source: 2018/19 World Bank Integrated Household Survey. All values in 2019 MWK
cost)3. This partly reflects the high labor requirements for tobacco farming. Additionally, cost to transport tobacco to buyers are also markedly higher than other alternative crops. This is likely due to the auction floors being the mandatory sales locations whereas the other crops could be sold locally. Overall, the high gross revenues of tobacco dominate its high costs resulting in strikingly higher returns to land4 and labor5 compared to its alternatives.
6. Policy Scenarios to improve competitiveness of tobacco alternatives
The results in the previous section indicated return to land for tobacco are much higher than that of alternatives. In order to identify the potential policy levers to raise the competitiveness of alternative crops we perform sensitivity analysis on returns to land of tobacco and alternative crops. Specifically, we calculate the change in returns to land
3 The median hours per day spent by own HH labor were same for crops under consideration (8 hours per
day). 4 Ret. to land (MWK/ha) = Gross Revenue(MWK/ha) ‒ Total costs excluding land costs(MWK/ha) 5 Ret. to labor(MWK/days) = {Gross Rev.(MWK/ha) ‒ Tot. costs excl. HH labor costs(MWK/ha)} / HH
labor(days/ha)
93.7
81.7
1.5
11.5
32.2
17.8
1.1
2.5
22.3
21.8
0.9
0.5
19.7
19.0
0.7
0.8
17.8
15.5
0.4
2.2
0.0 20.0 40.0 60.0 80.0 100.0
10% increase in Output Price
10 % increase in Yield
10% decrease in Seed price
10% decrease in Fertilizer price
Change in returns to land (000's MWK/ha) Pigeon pea Soya bean Groundnut Maize Tobacco
Shah et al.
MwAPATA Working Paper 21/03 18
Figure 6: Farm-gate tobacco price tax and price subsidy levels of alternative crops needed to be equally profitable as tobacco cultivation.
(Source: 2018/19 World Bank Integrated Household Survey. Note: The resulting real returns to land (000ʼs
MWK per ha) from balanced budget scenario are as follows: 83.5 (Soybean and Tobacco), 67.9 (Pigeon Pea
and Tobacco), 33.3 (Maize and Tobacco), and 5.0 (Ground nut and Tobacco). All values in 2019 Malawi Kwacha.
corresponding to a 10% change in four determinants of returns to land (fertilizer price, seed price, yield, and output price). Figure 5 shows that returns to land of all crops are more responsive to revenue increasing determinants (yield and output price) than to cost reducing determinants (fertilizer price and seed price). Thus, we explore policy scenarios that would increase the gross revenues of other crops and in turn bring the returns to land of alternatives closer to that of tobacco.
6.1 Output price improving scenarios:
Providing price subsidy is often suggested to improve the profitability of the crop. However, price subsidies can exert tremendous pressure on the budget. Hence, in order to relax the fiscal burden, we look at a balanced budget scenario where the price subsidy for the crops under investigation is fully funded through revenue generated from a tax on tobacco. The objective is to understand how much price subsidy would be needed for each crop so as to
-38%
-35%
-31%
-29%
69%
14%
144%
93%
-50% -30% -10% 10% 30% 50% 70% 90% 110% 130% 150%Tax Price Subsidy
Maize
GroundnutTobacco
Tobacco
Tobacco
Tobacco
Soya Bean
PigeonPea
Shah et al.
MwAPATA Working Paper 21/03 19
make its returns to land similar to that of tobacco subject to self-sufficiency constraint. In doing so, we assume no supply response to price changes and no transaction costs. Figure 6 shows the price subsidy of crops and corresponding tobacco tax needed to make the alternative as competitive as tobacco. We find that a 29 % tax on tobacco would be needed to support a 93% soybean subsidy. A 144% price subsidy for pigeon pea funded through 31% tobacco tax would make pigeon pea as competitive as tobacco. While the average returns to land of maize are higher than that of soybean and pigeon pea, a much higher tobacco tax would be required to fund a maize price subsidy since there are a large number of maize producers in Malawi. Lastly, a 69% price subsidy supported through a 38% tobacco tax would be required to make groundnut as competitive as tobacco. Furthermore, even higher levels of price subsidies would be required in absence of tobacco tax since tobacco returns to land are highly sensitive to changes in own price. Thus, government funded price supports with or without tobacco tax might not be feasible and there is need to explore new international and domestic markets for alternatives to raise their competitiveness compared to tobacco.
6.2 Yield improvement scenarios:
Another mechanism to increase the profitability of alternative crops to tobacco would be to improve yields of these crops. In terms of the potential for increasing maize yields, encouraging more fertilizer use and shifting from local seeds to improve maize seeds are often seen as ways to raise the yield levels of smallholder farmers. However, figure 7 shows that even with the consideration that all smallholders adopt improved maize varieties, fertilizer use would need to increase to an unrealistically high rate of 1.95 ton/ha in order for maize returns to land to match tobacco levels.6. This assumes an optimistic nitrogen use efficiency rate of 14 kilograms of maize per kilogram of nitrogen based on estimates from the Maize Productivity Task Force in Malawi from the mid 1990ʼs (MPTF, 1999). In fact, a
6 We use linear functional form to predict the maize yield response to fertilizer application. Local maize
response equation: Yield(kg/ha) = 1301.6022 + (7/3)*fertilizer applied (kg/ha). Improved maize response
equation: Yield(kg/ha) = 1512.8877 + (14/3)* fertilizer applied (kg/ha). While the linear response equations is
not appropriate at high levels of fertilizer use such as 1.95 tons/ha, it provides with an upper bound estimate
on yield response. The fertilizer use efficiency is calculated by dividing the nitrogen use efficiency (NUE) by 3.
Following (Snapp et al., 2014), the NUE for local maize was assumed to be 7 and NUE of improved maize seed
was assumed to be 14. Improved maize varieties include OPV, hybrid, and hybrid recycled varieties.
Shah et al.
MwAPATA Working Paper 21/03 20
Figure 7: Real returns to land in 2019 Kwacha from improved maize productivity scenario.
Source: 2018/19 World Bank Integrated Household Survey
recent review of NUE rates on Malawian smallholderʼs fields reveals that rates are actually much lower than estimate from the MPTF (Burke et al., 2021). This suggests that at the current efficiency of improved maize varieties and fertilizer it is unlikely to increase the return to land of maize to that of tobacco.
The second scenario presented in Figure 7 also indicates that at the government recommended fertilizer application rate of 250 kg/ha, the nitrogen use efficiency (NUE) of improve maize seeds would need to increase to 48 kilogram / ha (Snapp et al., 2014). This would require massive investments in complimentary inputs such as better soil fertility management, improved water control and more efficient use of labor in order to boost NUE. However, reaching an NUE of 48 is unrealistic in the average Malawian smallholder production system at any time in the near future.
Figure 8 show the yield increases required for soybean, pigeon pea and groundnut to raise their average returns to land to tobacco levels. Soybean and groundnut yields need to increase by around 200 % from their 2018/19 yield levels and pigeon pea yields would need to increase by more than 300%. Thus, yields of soybean, ground nut and pigeon pea need to be more than 3000 kg/ha to match tobacco returns to land. Furthermore, due to limited
359 359
0
100
200
300
400
Maize Maize
Re
turn
s to
la
nd
(0
00
’s M
WK
/ha
Tob. returns to land (357)
Improved Seeds
+
250 kg/ha fertilizer use
NUE : 48
Improved Seeds
+
1.95 tons/ha fertilizer use
NUE : 14
Shah et al.
MwAPATA Working Paper 21/03 21
Figure 8: Yield increases required for soybean, ground nut and pigeon pea to be equally profitable as tobacco.
Source: 2018/19 World Bank Integrated Household Survey
information on NUE for these crops, we do not take into account the cost of productivity improving technology ( i.e., seeds or fertilizer to realize these increased yields). Hence, these estimates should be considered as lower bound values. This scenario demonstrates that current levels of soybean, groundnut, and pigeon pea yields need to increase by a vast amount and underscores the need for research on development and adoption of improved seed varieties.
7. Conclusions
The analysis included in this report provides important insights into tobacco profitability for smallholder farmers over time and in relation to other alternative crops. Alternative crops include maize, groundnut, soybean, and pigeon pea. The main source of data on smallholder households came from the IHS and IHPS survey collected by the World Bank in 2009/10, 2012/13, 2015/16 and 2018/19. We also obtained data on tobacco prices over time in Malawi from the Tobacco Control Commission.
We found evidence that just 20% of households grew tobacco in one of the four IHPS survey waves grew it in every wave and only 16% of households grew tobacco in all four survey waves. Interestingly, 45% of households dropped tobacco after 2009/10, while 39% of households added tobacco cultivation in 2009/10. This suggests that most households
218%
226%
314%
0% 50% 100% 150% 200% 250% 300% 350%Change in yield
Pigeon Pea
Soya Bean
Groundnut
Shah et al.
MwAPATA Working Paper 21/03 22
are not wed to tobacco cultivation and adjust their crop based on changes in agronomic and market conditions across years. Those who left tobacco farming had less land and family labor compare to the households that grew tobacco.
Analysis of tobacco prices suggests that the real price of tobacco has declined over the past ten years in Malawi, but the downward trend in tobacco prices seems to be a reversion back to price levels from 15-20 years ago before the global commodity boom of the late 2000ʼs. The household level data is consistent with this as our results suggest that the number of smallholders growing tobacco declined from 15% in ʻ09/10 to about 5% in ʻ18/19. While the participation in tobacco declined over the years, average area under tobacco grew by 33% and average yields increase by 40% from 2009/10 to 2018/19. Moreover, for those who continued to cultivate tobacco, it remained profitable. This was true both in terms of gross revenue and returns to land and labor compared to alternative crops. Prices for maize, groundnut, soybean and pigeon pea never approached prices for tobacco at the farmgate and were often only 20-40% of tobacco prices.
Sensitivity analysis of cost of production suggests that tobacco returns to land are much more responsive to price changes than other crops. Thus, drop in prices of tobacco over the last ten years led the households with sufficient land and family labor to intensify tobacco cultivation to make up for the reduction in returns to land due to price change. The households that didnʼt have enough land and labor endowments moved out of tobacco farming to cultivate more maize and legumes. Overall, these findings suggest that the tobacco sub-sector in Malawi may have undergone something of a structural transformation over the past decade. Inefficient tobacco producers have left for other income generating activities as the price of tobacco declined, while the most efficient producers remained. As such, our policy scenarios clearly indicate that it will be very expensive to induce the remaining 5% of households to move away from tobacco through a combination of taxes on tobacco and subsidies for other crops. Therefore, we suggest that it would be more cost-effective for policy priorities in Malawi to focus on improving value chains for other high value crops rather than directly focusing on moving the remaining 5% of smallholder farmers out of tobacco cultivation.
Shah et al.
MwAPATA Working Paper 21/03 23
Acknowledgements
This research was produced with the help of a grant from the Foundation for a Smoke-Free World (FSFW) Agricultural Transformation Initiative (ATI) through the Michigan State University (MSU) Food Security Group. We are grateful for this generous support. Any views expressed herein are those of the authors and do not necessarily represent the views of the donors.
References
Alwang, J., & Siegel, P. B. (1999). Labor shortages on small landholdings in Malawi: Implications for
policy reforms. World Development, 27(8), 1461‒1475. https://doi.org/10.1016/S0305-
750X(99)00065-0
Burke, W. J., Snapp, S. S., Peter, B., & Jayne, T. S. (2021). Sustainable intensification in jeopardy:
transdisciplinary evidence from Malawi. Working paper
MPTF (Maize Productivity Task Force, A. G. I). (1999). Validating and strengthening the area-
specific fertilizer recommendations for hybrid maize grown by Malawian smallholders: A
research report of the results of the nationwide 1997/98 Maize Fertilizer Recommendations
Demonstration. Lilongwe, Malawi
Otañez, M. G., Mamudu, H., & Glantz, S. A. (2007). Global leaf companies control the tobacco
market in Malawi. Tobacco Control, 16(4), 261. https://doi.org/10.1136/tc.2006.019273
Ricker-Gilbert, J., Jumbe, C., & Chamberlin, J. (2014). How does population density influence
agricultural intensification and productivity? Evidence from Malawi. Food Policy, 48, 114‒128.
https://doi.org/10.1016/j.foodpol.2014.02.006
Snapp, S., Jayne, T. S., Mhango, W., Benson, T., & Ricker-Gilbert, J. (2014). Maize yield response to
nitrogen in Malawiʼs smallholder production systems. IFPRI Working Paper, October, 13.
http://ebrary.ifpri.org/utils/getfile/collection/p15738coll2/id/128436/filename/128647.pdf%0A
https://ageconsearch.umn.edu/record/188570
Xia, F., & Deininger, K. (2019). Spillover Effects of Tobacco Farms on the Labor Supply, Education,
and Health of Children: Evidence from Malawi. American Journal of Agricultural Economics,
101(4), 1181‒1202. https://doi.org/10.1093/ajae/aaz011
United Nations (2017). UN Comtrade database. New York, USA. https://comtrade.un.org/data
[27.09.2017]
Shah et al.
MwAPATA Working Paper 21/03 24
Appendix
Figure A1: Sensitivity of returns to land to change in output price
Source: 2018/19 World Bank Integrated Household Survey. All values in 2019 MWK
-93.7
93.7
-32.2
32.2
-22.3
22.3
-19.7
19.7
-17.8
17.8
-100.0
-80.0
-60.0
-40.0
-20.0
0.0
20.0
40.0
60.0
80.0
100.0
-10% 0% 10%
Ch
an
ge
in
re
turn
s to
la
nd
(0
00
's M
WK
/ha
)
Change in output price (%)
Tob. Maize G.Nut S. Bean P.Pea
Shah et al.
MwAPATA Working Paper 21/03 25
Figure A2: Sensitivity of returns to land to change in yield
Source: 2018/19 World Bank Integrated Household Survey. All values in 2019 MWK
-81.7
81.7
-17.8
17.8
-21.8
21.8
-19.0
19.0
-15.5
15.5
-100.0
-80.0
-60.0
-40.0
-20.0
0.0
20.0
40.0
60.0
80.0
100.0
-10% 0% 10%
Ch
an
ge
in
re
turn
s to
la
nd
(0
00
's M
WK
/ha
)
Change in yield(%)
Tob. Maize G.Nut S. Bean P.Pea
Shah et al.
MwAPATA Working Paper 21/03 26
Figure A3: Sensitivity of returns to land to change in seed price
Source: 2018/19 World Bank Integrated Household Survey. All values in 2019 MWK
1.5
-1.5
1.1
-1.1
0.9
-0.9
0.7
-0.7
0.4
-0.4
-2.0
-1.5
-1.0
-0.5
0.0
0.5
1.0
1.5
2.0
-10% 0% 10%
Ch
an
ge
in
re
turn
s to
la
nd
(0
00
's M
WK
/ha
)
Change in seed price(%)
Tob. Maize G.Nut S. Bean P.Pea
Shah et al.
MwAPATA Working Paper 21/03 27
Figure A4: Sensitivity of returns to land to change in fertilizer price
Source: 2018/19 World Bank Integrated Household Survey. All values in 2010 MWK
11.5
-11.5
2.5
-2.5
0.5-0.5
0.8
-0.8
2.2
-2.2
-15.0
-10.0
-5.0
0.0
5.0
10.0
15.0
-10% 0% 10%
Ch
an
ge
in
re
turn
s to
la
nd
(0
00
's M
WK
/ha
)
Change in fertilizer price(%)
Tob. Maize G.Nut S. Bean P.Pea
Shah et al.
MwAPATA Working Paper 21/03 28
Table A1: District level tobacco participation by smallholders over time.
Percentage of households in the
district growing tobacco District 2009/10 2015/16 2018/19 Dowa 52% 23% 16% Kasungu 46% 20% 18% Ntchisi 43% 22% 18% Rumphi 40% 27% 22% Lilongwe 27% 15% 10% Mchinji 27% 9% 11% Chitipa 24% 7% 5% Mzimba 21% 13% 12% Zomba 13% 3% 1% Dedza 8% 2% 2% Ntcheu 8% 2% 2% Mangochi 7% 0% 1% Machinga 7% 1% 1% Phalombe 7% 4% 7% Chiradzu 5% 1% 2% Balaka 5% 0% 1% Nkhotakota 5% 0% 1% Lilongwe City 4% 5% 2% Nkhata bay 4% 3% 3% Karonga 3% 0% 1% Salima 3% 1% 1% Mulanje 3% 0% 1% Blanytyr 1% 0% 0% Nsanje 0% 0% 0% Neno 0% 0% 0% Mwanza 0% 0% 0% Zomba City 0% 0% 1% Mzuzu City 0% 2% 1% Thyolo 0% 0% 0% Chikwawa 0% 0% 0% Blantyre 0% 0% 0% Likoma 0% 2% 0% Source: World Bank Integrated Household Survey
Shah et al.
MwAPATA Working Paper 21/03 29
Table A2: Plot-level participation rates (%) for various crops over time % of all plots 2009/10 2015/16 2018/19 MAIZE 71% 75% 69% NKHWANI 8% 25% 33% PIGEONPEA 19% 21% 24% GROUNDNUT 14% 9% 14% BEANS 6% 9% 11% SOYABEAN 3% 6% 7% SORGHUM 5% 7% 7% RICE 4% 4% 5% SWEET POTATO 2% 2% 4% TOBACCO 8% 4% 3% PEAS 1% 2% 2% SUNFLOWER 1% 1% 1% PEARL MILLET 1% 1% 1% THERERE/OKRA 0% 0% 1% IRISH POTATO 1% 0% 1% FINGER MILLET) 1% 0% 1% COTTON 2% 2% 1% TOMATO 0% 0% 1% GROUND BEAN 0% 0% 1% SUGAR CANE 0% 0% 0% TANAPOSI 0% 0% 0% ONION 0% 0% 0% PAPRIKA 0% 0% 0% CABBAGE 0% 0% 0% WHEAT 0% 0% 0%
Source: World Bank Integrated Household Survey
Table A3: Participation rates (%) for various crops amongst households that stop growing tobacco and households that grow tobacco. 2012/13 2015/16 2018/19
Percentage growing each
crop
HHs that stopped
growing tob. in 2012/13 but
grew it in 2009/10
HHs that grew tob.
in 2012/13
(1) - (2)
HHs that stopped
growing tob. in 2015/16 but grew it in 2012/13
HHs that grew tob.
in 2015/16
(4) - (5)
HHs that stopped
growing tob. in 2018/19 but
grew it in 2015/16
HHs that grew tob. in
2018/19
(7) - (8)
(1) (2) (3) (4) (5) (6) (7) (8) (9) MAIZE 90 99 -9 83 98 -16 80 96 -16 TOBACCO 0 100 -100 0 100 -100 0 100 -100 GROUNDNUT 51 63 -12 20 37 -17 39 46 -7 RICE 2 2 0 0 0 0 2 1 1 GROUND BEAN 2 1 0 0 0 0 1 1 0 SWEET POTATO 5 6 -1 4 5 -1 9 8 2 IRISH POTATO 2 1 2 1 0 1 1 5 -4 FINGER MILLET 1 0 1 2 0 2 1 4 -3 SORGHUM 5 11 -6 10 8 1 9 12 -3 PEARL MILLET 0 2 -2 2 3 -1 1 0 1 BEANS 12 23 -11 8 10 -2 18 29 -11 SOYABEAN 17 23 -7 16 21 -5 24 27 -3 PIGEONPEA 16 20 -4 13 10 3 15 18 -3 COTTON 3 1 2 3 0 3 0 1 -1 SUNFLOWER 1 1 0 1 1 0 1 2 -1 SUGAR CANE 0 0 0 0 0 0 0 1 -1 CABBAGE 1 0 1 0 0 0 0 0 0 TANAPOSI 1 0 1 0 2 -2 1 1 0 NKHWANI 16 37 -22 17 27 -10 52 58 -6 THERERE/OKRA 0 4 -4 1 2 -1 0 1 -1 TOMATO 1 1 0 0 4 -4 1 0 1 ONION 1 0 1 0 0 0 0 0 0 PEAS 4 9 -5 2 3 -1 1 0 1 PAPRIKA 0 0 0 0 0 0 1 0 1
Source: World Bank Integrated Household Survey
Mw
APAT
A W
orki
ng P
aper
21/
03
16
Shah
et a
l.